Portworx Volume backup to Cloud

I am looking for backup solution using Portworx for my Kubernetes cluster. I use IBM Cloud’s IKS(Kubernetes service) and they provide Portworx Enterprise with IKS. But I have an issue using object store for my backup. Can I use a high performance backup target other than object store with Portworx?

Hi @Sandeepkallazhi as far as I know, cloud backups must be s3 compatible…I was in IBM but never worked with the storage, and following the documentation, I saw compatibility with azure, gcs and s3. Not an expert, but I think you can install minio and use it as target storage.
Minio: https://www.katacoda.com/portworx/scenarios/px-minio
This is my point of view from user, maybe a PX expert can give you a better answer.

1 Like

You are correct Azure, Google and Amazon S3 are supported. As mentioned S3 Compatible is also supported which means Minio and other storage vendors that support S3 Compatibility can also be used. For instance Minio, Ceph, Wasabi, Cloudian all have S3 compatibility. I would look into if your target vendor supports an S3 compatible endpoint


@Gerardo_Hernandez @Ryan_Wallner I have tried backing up using Minio and it works fine. But then I would need to build it to scale for HA, for example use a distributed Minio deployment.
On a side note why is that all tools which are in the space, i.e for back up of K8s like Velero, Kasten and Portworx all require object store? Is it because K8s itself is a bunch of objects? But then why not have a better backup target for the volumes of the stateful apps that run on K8s.

Backups are categorized as a “cold” data type meaning they are access quite infrequently. Object storage offers a scalable and very cheap (keyword) secondary storage solution compared to faster primary storage. This makes object storage optimal for backups, let alone it is industry standard to use it. Often these solutions are already massively scalable and reliable such as amazon S3, minio and ceph are some of the exceptions as you would need to run them yourself. This is likely why you see many solutions and products in this space use object storage endpoints especially the big ons from google, azure, and aws

Curious what “better backup target” you would like to see used for these types of solutions and what use case you had in mind was as we are always interested in feedback.

thanks for your response!